UNC-51/ATG1 kinase regulates axonal transport by mediating motor ...

Download PDF
2 downloads 27 Views 1MB Size Report
Comment
1Division of Neurosciences, Beckman Research Institute of the City of Hope, ... of Richmond, Virginia 23173, USA; 4Beckman Laser Institute, University of ...
UNC-51/ATG1 kinase regulates axonal transport by mediating motor–cargo assembly Hirofumi Toda,1,2,6 Hiroaki Mochizuki,2,6 Rafael Flores III,1 Rebecca Josowitz,3 Tatiana B. Krasieva,4 Vickie J. LaMorte,4 Emiko Suzuki,5 Joseph G. Gindhart,3 Katsuo Furukubo-Tokunaga,2,8 and Toshifumi Tomoda1,7 1

Division of Neurosciences, Beckman Research Institute of the City of Hope, California 91010, USA; 2Graduate School of Life and Environmental Sciences, University of Tsukuba, Tsukuba 305-8572, Japan; 3Department of Biology, University of Richmond, Virginia 23173, USA; 4Beckman Laser Institute, University of California at Irvine, Irvine, California 92697, USA; 5Gene Network Laboratory, National Institute of Genetics, Mishima, Shizuoka 411-8540, Japan

Axonal transport mediated by microtubule-dependent motors is vital for neuronal function and viability. Selective sets of cargoes, including macromolecules and organelles, are transported long range along axons to specific destinations. Despite intensive studies focusing on the motor machinery, the regulatory mechanisms that control motor–cargo assembly are not well understood. Here we show that UNC-51/ATG1 kinase regulates the interaction between synaptic vesicles and motor complexes during transport in Drosophila. UNC-51 binds UNC-76, a kinesin heavy chain (KHC) adaptor protein. Loss of unc-51 or unc-76 leads to severe axonal transport defects in which synaptic vesicles are segregated from the motor complexes and accumulate along axons. Genetic studies show that unc-51 and unc-76 functionally interact in vivo to regulate axonal transport. UNC-51 phosphorylates UNC-76 on Ser143, and the phosphorylated UNC-76 binds Synaptotagmin-1, a synaptic vesicle protein, suggesting that motor−cargo interactions are regulated in a phosphorylation-dependent manner. In addition, defective axonal transport in unc-76 mutants is rescued by a phospho-mimetic UNC-76, but not a phospho-defective UNC-76, demonstrating the essential role of UNC-76 Ser143 phosphorylation in axonal transport. Thus, our data provide insight into axonal transport regulation that depends on the phosphorylation of adaptor proteins. [Keywords: Axonal transport; unc-51; kinesin adaptor; phosphorylation; motor–cargo assembly] Supplemental material is available at http://www.genesdev.org. Received March 10, 2008; revised version accepted October 10, 2008.

The axon serves as a track for long-distance transport of synaptic components that are synthesized within cell soma and delivered to nerve terminals (for reviews, see Brady 1991; Vallee and Sheetz 1996; Guzik and Goldstein 2004). The microtubule (MT) transport system supports most axonal traffic, in which kinesins are the major anterograde motors responsible for the transport of a wide range of cargoes (i.e., synaptic vesicles [SVs], mitochondria, cytoskeletal elements, and mRNAs), whereas dyneins are responsible for the retrograde transport of cargoes (Hirokawa 1998; Vale 2003). A number of transport components have been identified, and the recent discovery of a large family of structurally related motor proteins has addressed the question

6 These authors contributed equally to this work. Corresponding authors. 7 E-MAIL [email protected]; FAX (626) 446-5774. 8 E-MAIL [email protected]; FAX 81-298-53-6644. Article is online at http://www.genesdev.org/cgi/doi/10.1101/gad.1734608.

3292

of motor–cargo specificity (Miki et al. 2005). The kinesin superfamily proteins (KIFs) are comprised of >30 members, and each KIF is responsible for carrying a specific subset of cargoes. In addition, protein–protein interaction screens with several motor proteins have identified multiple adaptor proteins, which further determine the specificity of cargoes transported by each motor (Guzik and Goldstein 2004; Hirokawa and Takemura 2005). For example, Drosophila kinesin heavy chain (KHC), the catalytic component of kinesin-1, binds an adaptor, Milton, that specifically recruits mitochondria as cargo and transports them toward synapses (Stowers et al. 2002). KHC also binds UNC-76/FEZ1, an evolutionarily conserved adaptor protein (Bloom and Horvitz 1997; Kuroda et al. 1999; Blasius et al. 2007), and loss of unc-76 results in defective SV transport in Drosophila segmental nerves (SGNs) (Gindhart et al. 2003). Kinesin light chain (KLC) interacts with a scaffolding protein c-Jun N⬘-terminal kinase-interacting protein 1 (JIP1) (Verhey et al. 2001; Horiuchi et al. 2005) or an additional scaffolding protein

GENES & DEVELOPMENT 22:3292–3307 © 2008 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/08; www.genesdev.org

Regulation of axonal transport by unc-51 kinase

UNC-16/SYD/JIP3 (c-Jun N⬘-terminal kinase-interacting protein 3) in Caenorhabditis elegans, Drosophila, and mammals (Bowman et al. 2000; Byrd et al. 2001; Verhey et al. 2001). JIP1 in turn binds amyloid precursor protein (APP) (Scheinfeld et al. 2002), a kinesin-1 cargo that is implicated in normal axonal transport and in Alzheimer’s disease (Torroja et al. 1999; Kamal et al. 2001; Lazarov et al. 2005). Despite this progress in understanding motor–cargo specificity, it is not well understood how motor–cargo assembly is regulated during transport. Many components of the axonal transport machinery, such as kinesins, are phosphoproteins (Hollenbeck 1993), suggesting that phosphorylation may play a role in axonal transport. Recent reports have addressed the roles of glycogen synthase kinase 3␤ (GSK3␤), c-Jun N⬘-terminal kinase (JNK) and Ca2+-calmodulin kinase II (CaMKII), which negatively impact motor–cargo assembly during transport (Morfini et al. 2002; Horiuchi et al. 2007; Guillaud et al. 2008). However, little is known about the mechanisms of how organelle motility is dynamically controlled or maintained in vivo. Members of the conserved Ser/Thr kinase UNC-51 family (unc-51/Unc51.1/Unc51.2) are key regulatory genes that control axonal elongation during nervous system development in C. elegans and mice (Hedgecock et al. 1985; Ogura et al. 1994; Tomoda et al. 1999; Zhou et al. 2007). Murine Unc51.1 is expressed in a number of neuronal populations during development, including cerebellar granule cells, in which Unc51.1 is localized to both elongating axonal shafts and growth cones. Inhibition of Unc51.1 activity leads to retarded neurite extension and parallel fiber formation in cerebellar granule neurons (Tomoda et al. 1999). A recent study showed that axon formation is regulated by the endocytic membrane traffic pathway via an Unc51.1-containing protein complex (Unc51.1/SynGAP/Rab5/Syntenin) (Tomoda et al. 2004). UNC-51 also interacts with VAB-8, which contains a kinesin-motor like domain, and these proteins cooperatively regulate axon extension in C. elegans (Lai and Garriga 2004). In addition, UNC-14, a protein that interacts with UNC-51 (Ogura et al. 1997), plays a role in kinesin-1-dependent axonal transport in C. elegans (Sakamoto et al. 2005). Furthermore, unc-51 mutants in C. elegans manifest abnormal accumulation of intrasomal vesicles that are positive for UNC-5, a Netrin/ UNC-6 receptor (Ogura and Goshima 2006), suggesting a role for UNC-51 in anterograde vesicle transport. Collectively, these lines of evidence suggest that UNC-51 plays a role in intracellular transport. In this study, we demonstrate a role for UNC-51-mediated phosphorylation in the regulation of axonal transport in vivo. Loss of unc-51 results in defective axonal transport in a cargo type-dependent manner. UNC-51 binds and phosphorylates the kinesin adaptor UNC-76. When phosphorylated at Ser143, UNC-76 has an elevated affinity for Synaptotagmin-1 (Syt-1), a major transmembrane protein of SVs (Südhof 2004), whereas non-phospho-UNC-76 fails to bind Syt-1. Characterization of unc-51 therefore reveals a key step in the regulation of

motor–cargo interaction and provides insight into the dynamic control of axonal transport in vivo.

Results Generation of unc-51-null mutants in Drosophila The Ser/Thr kinase UNC-51 is implicated in axonal development in C. elegans and mice (Ogura et al. 1994; Tomoda et al. 1999; Zhou et al. 2007). The Drosophila genome has a single unc-51 homolog (CG10967), which is also referred to as atg1 (Scott et al. 2007) or Pegarn (Ahantarig et al. 2008). Expression analysis of unc-51 mRNA in situ in Drosophila revealed that it was widely distributed throughout stage 2 embryos, demonstrating a high level of maternal expression (Supplemental Fig. S1A). The unc-51 mRNA continued to be expressed throughout the embryonic stages, and was progressively restricted to the nervous system, suggesting a role for unc-51 in neuronal development. To further elucidate the neuronal functions of unc-51, we generated null alleles in Drosophila. P-element EP(3)3348 is a semilethal insertion at adult stage and is inserted within the first exon of the unc-51 gene, 539 base pairs (bp) upstream of the ATG initiation codon (Fig. 1A). Precise excision of this P-element rescued the semilethality (data not shown). Two unc-51 mutant alleles, unc-513 and unc-5125, were generated by imprecise excision, which deleted 804 bp and 1787 bp, respectively, downstream from the insertion point (Fig. 1A), removing part of the first exon, including the ATG initiation codon. These mutations caused axonal transport defects at the larval stage, as well as pupal lethality, and failed to complement each other. Genetic crosses of unc-513 or unc-5125 with an unc-51 deficiency line [unc-513/Df(3L)ED4486 or unc-5125/Df(3L)ED4486] gave rise to phenotypes identical to unc-513/unc-5125, displaying both defective axonal transport and pupal lethality. Immunohistochemical analyses confirmed loss of UNC-51 protein expression in the mutants (Supplemental Fig. S1B). Thus, the evidence indicates that both alleles are null for the unc-51 locus.

Phenotypic analysis of axon development in unc-51 mutants Consistent with previous studies on the role of unc-51 in axon formation in worms and mice (Ogura et al. 1994; Tomoda et al. 1999), Drosophila unc-51 mutants exhibited varying degrees of defective axonal tracts in the ventral nerve cord (VNC) at the embryonic stage, including premature truncation and abnormal midline crossing of longitudinal tracts (H. Mochizuki, H. Toda, T. Tomoda, and K. Furukubo-Tokunaga, unpubl.). Approximately 47% of the mutants were estimated to be dead by the early third instar larval stage, likely due to defective axon formation during the embryonic period. In contrast, the remainder (∼53%) survived through the larval stage, with the SGNs of third instar larvae extended (Supple-

GENES & DEVELOPMENT

3293

Toda et al.

Figure 1. unc-51 functions in axonal transport. (A) Molecular structures of the unc-51-null mutant alleles. P-element EP(3)3348 is inserted within the first exon of the unc-51 gene, 539 bp upstream of the initiation ATG codon. Two unc-51-null alleles, unc-513 and unc-5125, were generated by imprecise excision. PCR amplification of these unc-51 alleles (blue arrows indicate locations of primers) and subsequent sequencing analyses confirmed that unc-513 has an 804 bp deletion and unc5125 has a 1787-bp deletion. Both alleles lack the start codon, and did not affect the expression unit of CG17667 located within intron 2. (B) SGNs of unc-51-null mutant (unc-513/unc-5125) or wild-type third instar larvae immunostained with anti-Syt-1 (red) and anti-CSP (green). Syt1-positive aggregates colocalized with those positive for CSP in unc-51 mutants. Bar, 10 µm. (C) Ultrastructure of SV aggregates in unc-51 mutants. Electron micrograph of a cross-section of an unc-51 mutant SGN immunostained with anti-Syt-1. Arrowheads indicate SVs positive for antiSyt-1. (Table) The number of SV aggregates (defined as a cluster of >20 SVs per aggregate) was scored in electron-micrographs from unc-51 mutants (n = 34) and wild type (n = 23). Bar, 200 nm. (D) Genetic rescue of axonal transport defect in unc-51 mutants by unc-51 transgene. CSP-positive SV aggregation phenotype in unc-51 mutants was rescued by pan-neuronal expression of unc-51/wild-type transgene (elavc155 > UAS-unc-51WT; unc-513/unc5125), but not by a kinase-deficient unc-51 (elavc155 > UAS-unc-51K38A; unc-513/unc5125). Numbers of CSP-positive puncta per 50 µm SGN (mean ± SEM) were plotted for each genotype (graph). (**) Statistically significant rescue (P < 0.01, Student t test). (n.s.) Not significant (P = 0.843). Bar, 10 µm.

mental Fig. S1C) and connected to normal muscle targets (data not shown), and died during the pupal stage. Electron microscopy (EM) analysis of SGN cross-sections at the third instar stage revealed that the number of axons per SGN and the average axon caliber in mutant larval SGNs was indistinguishable from wild type (Supplemental Fig. S1D), suggesting that unc-51 mutants that survive to larval stage have little anatomical abnormality in axon formation. Selective membrane localization defects in unc-51 mutant larvae Because recent studies in worms and mice suggested a role for unc-51-mediated membrane organization in axon formation (Tomoda et al. 2004; Ogura and Goshima 2006), we tested several membrane markers to identify

3294

GENES & DEVELOPMENT

potential membrane defects within extended SGNs of unc-51 mutant third instar larvae. We observed aberrant accumulation of SVs within mutant SGNs, as evidenced by prominent aggregation of Syt-1 (Fig. 1B). This elevated level of Syt-1 in mutant SGNs correlated with fewer numbers of synaptic boutons, less intense Syt-1 immunostaining, reduced bouton area, and reduction in total amount of Syt-1 at neuromuscular junctions (NMJ) (segment A2, muscle 6/7) as compared with the wild type (Supplemental Fig. S2). An additional SV marker, cysteine string protein (CSP) (Zinsmaier et al. 1994), confirmed the phenotype by showing a nearly complete overlap in localization with Syt-1-positive aggregates in unc-51 mutants (Fig. 1B). Immuno-EM revealed that Syt-1-positive aggregates were comprised of clusters of SVs within the axon shaft (Fig. 1C). The SV transport defect in unc51 mutants was rescued by unc-51 transgene expression

Regulation of axonal transport by unc-51 kinase

(elavc155 > UAS-unc-51;unc-513/unc-5125), attributing the mutant phenotype to the loss of unc-51 activity (Fig. 1D). A kinase-deficient unc-51 transgene carrying a point mutation in the ATP-binding site failed to rescue the mutant phenotype (elavc155 > UAS-unc-51K38A;unc-513/unc-5125), demonstrating a critical role for the kinase activity in organizing axonal membranes (Fig. 1D). To further address the selectivity of membrane types affected by unc-51 mutation, additional markers were tested. Rab5, an early endosomal marker, showed mild aggregates within mutant SGNs, in a pattern distinct from those of CSP-positive SVs (Supplemental Fig. S3A). A late endosome–lysosome marker, LAMP1, revealed many aggregates, some of which colocalized with SV aggregates (Supplemental Fig. S3B). In contrast, overall mitochondrial localization appeared unaffected in mutants and rarely overlapped with Syt-1 (Supplemental Fig. S3C), suggesting that SV aggregates do not serve as physical barriers for mitochondrial transport and that SV and mitochondria have distinct transport pathways. Quantitative analysis confirmed that the overall distribution of the sizes of mito⬋GFP-positive puncta smaller than 20 µm2 were equivalent between wild type and unc51 mutants (Supplemental Fig. S3D), although more large puncta (>20 µm2) were present in unc-51 mutants (Supplemental Fig. S3E), suggesting that mitochondrial transport is also affected in unc-51 mutants. Indeed, time-lapse video microscopic analysis of mitochondria labeled with mito⬋GFP (OK6 > UAS-mito⬋GFP) revealed that mitochondria in wild type moved at an average speed of 0.25 ± 0.02 µm/sec (n = 27), whereas those in unc-51 mutants moved at 0.16 ± 0.02 µm/sec on average (n = 36) (Supplemental Movie S1). Thus, the unc-51 mutation results in a varying degree of axonal membrane defects in a cargo type-dependent manner. In EM analysis, we observed clusters of a variety of membrane components within unc-51 mutant axons more frequently than in wild type. Those include dense core vesicles, multivesicular bodies (Supplemental Fig. S1E) and small clear vesicles that were reminiscent of SVs (Fig. 1C). Taken together, these results suggest that unc-51 is important for organizing axonal membranes in the larval SGNs.

unc-51 functions in SV transport The SV aggregation in unc-51 mutants (Fig. 1B) was reminiscent of that reported for mutants of Kinesin heavy chain (Khc) (Hurd and Saxton 1996), Kinesin light chain (Klc) (Gindhart et al. 1998), and Dynein heavy chain (Dhc) (Martin et al. 1999). As in these mutants, unc-51 mutants displayed sluggish larval crawling and occasional tail flipping (data not shown), a phenotype consistent with defective axonal transport. SVs are primarily generated in neuronal soma and delivered to synapses via anterograde transport machinery. To visualize SV transport defects in unc-51 mutant SGNs, timelapse video microscopy was performed (Supplemental Movie S2). In wild-type third instar larvae (OK6 > UASSyt-1⬋eGFP), most vesicles labeled with Syt-1⬋eGFP moved at velocities typical of normal axonal transport (as fast as 2.26 µm/sec; average 0.89 ± 0.06 µm/sec, n = 47) with 53.2% of vesicles moving anterogradely and 46.8% moving retrogradely, in good agreement with a recent study (Barkus et al. 2008). In contrast, prominent accumulations of Syt-1⬋eGFP-positive vesicles were observed in unc-51 mutants, and ∼95% of small individual puncta outside aggregates were immobile or moved slower than 0.2 µm/sec during the observation period (30 sec). Less than 5% of the Syt-1⬋eGFP-positive vesicles were mobile and the average speed of SV transport in unc-51 mutants was 0.05 ± 0.01 µm/sec (n = 62) (Supplemental Movie S2). Therefore, unc-51 plays a significant role in SV transport at larval stages.

unc-51 genetically interacts with components of kinesin-1-dependent axonal transport Given that unc-51 functions in SV transport, we investigated a potential genetic interaction of unc-51 with several genes previously implicated in axonal transport by testing for SV transport defects. Among those, unc-76 (Gindhart et al. 2003) and Klc showed a genetic interaction with unc-51 in a double heterozygous scheme (unc76+/−;unc-51+/− or Klc+/−,unc-51+/−) (Fig. 2; Supplemental Fig. S4), a condition often used to identify functional interactions among transport genes (Martin et al. 1999).

Figure 2. unc-51 genetically interacts with unc-76 in axonal transport. SGNs of wild type or the indicated mutants immunostained with anti-CSP at the third instar larval stage. Arrows show CSP-positive aggregates. Bar, 20 µm. The number of aggregates found per 50 µm of nerve shaft was scored for each genotype (graph). Statistical significance evaluated by Student t test. (**) P < 0.01. Error bars show ±SEM. Genotypes are w- (A), unc-513/+ (B), unc-76Df(1)107/+ (C), and unc76Df(1)107/+;unc-513/+ (D).

GENES & DEVELOPMENT

3295

Toda et al.

UNC-76 is a KHC adaptor protein that is necessary for kinesin-1-mediated axonal transport (Gindhart et al. 2003), and KLC is an accessory component of kinesin-1 (Gindhart et al. 1998). Quantitative analysis revealed that unc-76+/−;unc-51+/− double heterozygotes had more SV aggregates (∼7/50 µm) than unc-76+/− or unc-51+/− single heterozygotes ( unc76WT and unc-76−/−;elav > unc-76SD), the numbers of synaptic boutons, size of the bouton area, and total amount of CSP staining in NMJs were elevated as compared with nonrescued larvae (unc-76−/−;elav > unc76SA), suggesting that the UNC-76/wt or UNC-76/ S143D transgenes rescued the SV transport defect. Collectively, these data demonstrate a critical role for UNC51-mediated phosphorylation of UNC-76 Ser143 in controlling axonal transport.

Phosphorylation-dependent interaction of UNC-76 and Syt-1 Because loss of UNC-76 phosphorylation led to defective axonal transport (Fig. 5E) and loss of unc-76 resulted in

GENES & DEVELOPMENT

3297

Toda et al.

Figure 4. Disorganized localization of SV and motor complexes in unc-51 mutants. (A) SGNs of unc-51 mutants (unc-513/unc-5125) or wild-type third instar larvae immunostained with anti-UNC-76 (red) and anti-Syt-1 (green). Red arrows and arrowheads point to aberrant localization of the indicated proteins. The right panel shows enlarged image (arrowheads, UNC-76; arrows, Syt-1). The numbers of puncta positive for UNC-76 and/or Syt-1 in unc-51 mutant SGNs were scored and the percentages of single-positive or double-positive puncta are shown in the graph. SGNs are oriented from the proximal side (top) toward the distal (bottom). (B) SGNs of unc-51 mutants or wild-type larvae immunostained with anti-UNC-76 (red) and anti-KHC (green). Red arrows point to aberrant localization of the indicated proteins. The right panel shows enlarged image (asterisks, KHC/UNC-76 double-positive aggregates). The numbers of puncta positive for UNC-76 and/or KHC in unc-51 mutant SGNs were scored and the percentages of single-positive or double-positive puncta are shown (graph). (C) SGNs of unc-51 mutants or wild-type larvae immunostained with anti-CSP (red) and anti-KHC (green). Red arrows and arrowheads point to aberrant localization of the indicated proteins. Right panel shows enlarged image (arrowheads, CSP; arrows, KHC). (Graph) The numbers of puncta positive for CSP and/or KHC in unc-51 mutant SGNs were scored and the percentages of single-positive or double-positive puncta are shown. (D) SGNs of unc-51 mutant or wild-type larvae immunostained with anti-CSP (red) and anti-UNC-76 (green) and observed with confocal microscopy (with 100× objective; single optical section taken at 0.7-µm interval). SGNs are oriented from the proximal side (left) toward the distal (right). Areas positive for CSP, UNC-76 or both signals were calculated based on the confocal data using an image analysis program, and schematically represented on the graph. Bars, 10 µm.

3298

GENES & DEVELOPMENT

Regulation of axonal transport by unc-51 kinase

Figure 5. Phosphorylation of UNC-76 Ser143 by UNC-51 is necessary for axonal transport. (A) In vitro phosphorylation of UNC-76 by UNC-51 kinase. HA-tagged UNC-51 or UNC-51/K38A expressed in HEK293T cells was immunoprecipitated and incubated with His-UNC-76 in the presence of Mg2+ and [␥-32P]-ATP. Phosphoproteins were analyzed by autoradiograph. Black arrow indicates UNC-51 autophosphorylation. Red arrowhead indicates UNC-76 phosphorylation. (B) Myc-tagged UNC-76 and HA-tagged UNC-51 were coexpressed in HEK293T cells. Cell lysates were incubated for an indicated period of time with (+) or without (−) alkaline phosphatase (CIAP) and analyzed by immunoblot using anti-myc. Red bracket indicates mobility shifts of UNC-76. Black bracket indicates nonshifted UNC-76. (C) Myc-tagged wild type and Ser/Thr → Ala mutant UNC-76 were expressed in HEK293T cells with UNC-51 or UNC-51/K38A. Cell lysates were analyzed by immunoblot using anti-myc. Red bracket indicates mobility shifts of UNC-76. (D) Extracts from wild-type and unc-51 mutant larvae were analyzed by immunoblot using an anti-phospho-UNC-76 antibody. Anti-actin and anti-UNC-76 were used as loading controls. (E) Genetic rescue of axonal transport defect in unc-76 mutants by unc-76 transgene. CSP-positive aggregation (arrows) and KHC aggregation (arrowheads) in unc-76 mutants were rescued by unc76WT (top) or unc-76S143D (middle), but not by unc-76S143D transgene (bottom). Transgene expression was driven by elav. unc-76DF(1)107 hemizygous males were used as unc-76-null mutants. Third instar larvae for each genotype were stained with anti-KHC (green), anti-UNC-76 (red), and anti-CSP (blue). Note the equivalent expression levels of UNC-76 transgenes in VNC and SGN. (**) Statistical significance (P < 0.01, Student t test). (n.s.) Not significant (P = 0.18). Bars: red, 50 µm; white, 20 µm.

disorganized motor–cargo distribution (Supplemental Fig. S7), we hypothesized that the phosphorylation of UNC-76 by UNC-51 may regulate the affinity of UNC76 to SV cargoes. Although FEZ1, a rat homolog of UNC76, bound Syt-1 in a yeast two-hybrid screen (Bloom and Horvitz 1997), whether they physically interact was not successfully confirmed. Indeed, we did not detect an in-

teraction between Syt-1 and UNC-76 in coimmunoprecipitation assays using the HEK293T expression system (Fig. 6A). However, UNC-76 copurified with Syt-1 in the presence of UNC-51, but not in the presence of kinasedeficient UNC-51/K38A (Fig. 6A). The association between UNC-76 and Syt-1 was abolished by phosphatase treatment (Fig. 6B), indicating that their interaction is

GENES & DEVELOPMENT

3299

Toda et al.

phosphorylation-dependent. Although Syt-1 often binds its partners in a Ca2+-dependent manner (Südhof 2004), the interaction between Syt-1 and UNC-76 was Ca2+independent (Fig. 6C). In addition, the phosphomimetic

UNC-76/S143D associated with Syt-1, whereas the phosphodefective UNC-76/S143A did not (Fig. 6A). A direct interaction between UNC-76 and Syt-1 was further confirmed by Förster resonance energy transfer

Figure 6. UNC-51 regulates the affinity between UNC-76 and Syt-1. (A) UNC-51-mediated association of Syt-1 and UNC-76. Mycor HA-tagged UNC-51 (full-length), UNC-76 (full-length), and Syt-1 (full-length) were coexpressed in HEK293T cells as indicated. Cell lysates were immunoprecipitated by anti-HA followed by immunoblot with anti-myc. (B) Phosphorylation-dependent association of Syt-1 and UNC-76. HA-UNC-76, myc-Syt-1, and myc-UNC-51 or UNC-51/K38A were expressed in HEK293T cells as indicated. Cell lysates were immunoprecipitated with anti-HA. The immunecomplexes were incubated in the presence (+) or absence (−) of alkaline phosphatase (CIAP) and analyzed by immunoblot using anti-myc. (C) Syt-1–UNC-76 interaction does not require Ca2+. HA-UNC-76, myc-Syt-1, and myc-UNC-51 or UNC-51/K38A were expressed in HEK293T cells as indicated. Cell lysates were immunoprecipitated with anti-HA. The immune complexes were incubated in the presence of EGTA or CaCl2 and analyzed by immunoblot using anti-myc. (D) Syt-1–UNC-76 interaction detected by FRET. COS7 cells were transfected with Syt-1-CFP, YFP-UNC-76, YFP-UNC-76/S143D, YFP-UNC-76/S143A, or UNC-51/K38A in combination as indicated. For each combination, initial (red) and fluorescence after acceptor photobleaching spectra (green) were plotted for a range of wavelengths indicated. Increase in donor fluorescence after acceptor bleaching indicates FRET (green arrows). Quantification of FRET (graph). (**) Statistical significance (P < 0.01). (E) KHC/UNC-76/Syt-1 complex formation. GST alone (G) and GST fused with KHC (amino acids 675–975) (GK) were produced in E. coli, purified on Glutathione-Sepharose beads, and incubated with myc-UNC-76 and myc-Syt-1 expressed in HEK293T cells. Proteins that bound to the beads were eluted and analyzed by immunoblot using anti-myc.

3300

GENES & DEVELOPMENT

Regulation of axonal transport by unc-51 kinase

(FRET) analysis in cultured cells. Coexpression of Syt-1CFP and YFP-UNC-76 in COS7 cells gave rise to FRET between the two proteins (Fig. 6D), presumably via the phosphorylation of UNC-76 by endogenous UNC-51 kinase. Indeed, the FRET signal was attenuated by kinasedeficient UNC-51/K38A (Fig. 6D), suggesting a specific role of UNC-51 kinase activity in the Syt-1–UNC-76 interaction. In addition, Syt-1-CFP and the phosphomimetic YFP-UNC-76/S143D produced FRET, whereas Syt-1-CFP and the phosphodefective YFP-UNC-76/S143A did not (Fig. 6D), further confirming the role of Ser143 phosphorylation in the Syt-1–UNC-76 interaction. Thus, the data demonstrate that the adaptor protein UNC-76 directly interacts with Syt-1 on the SVs, and that this interaction is regulated by UNC-51 kinase activity. To further address whether KHC/UNC-76-containing motor complex could interact with Syt-1 in a phosphorylation-dependent manner, we carried out a GST pulldown assay (Fig. 6E). GST-fused KHC, which bound UNC-76 (Gindhart et al. 2003), interacted with Syt-1 in the presence of UNC-76/S143D, but not in the presence of UNC-76/S143A or unphosphorylated UNC-76/wt. Consistent with in vivo observation (Fig. 4B), UNC-76 bound KHC regardless of the phosphorylation status at Ser143 (Fig. 6E; Supplemental Fig. S10). Taken together, association of UNC-76 with Syt-1, but not with KHC, is mediated by phosphorylation of UNC-76 Ser143 by UNC51 kinase.

Discussion Nature of defective axonal transport in unc-51 mutants This study demonstrates that loss of unc-51 function affects the transport of several axonal cargoes, including SV and mitochondria. In unc-51 mutants, SV transport is severely attenuated and many SV aggregates are found along the larval SGN axons. SV aggregation in unc-51 mutants is similar to that observed in Khc mutants (Hurd and Saxton 1996). However, unlike Khc mutants, SV aggregates in unc-51 mutants do not contain mitochondria, suggesting that the aggregated SVs in unc-51 mutants do not cause overall “steric hindrance,” in which the impaired cargo interferes with the transport of other cargoes as a secondary effect. In unc-51 mutants, Rab5-positive membranes also exhibited a pattern of aggregation different from that for SVs, supporting the idea that loss of unc-51 results in defective axonal transport in a membrane type-dependent manner. This view is further supported by a recent analysis of unc-51 mutants in C. elegans that showed that only a subset of cargoes are selectively mislocalized, whereas a majority of other cargoes are not affected (Ogura and Goshima 2006). SVs are one of the most severely affected axonal cargoes in unc-51 mutants. Two anterograde motors, kinesin-1 and kinesin-3, have been implicated in SV transport in Drosophila. A recent study addressed an essential role of unc-104 (imac, kinesin-3) in SV transport (Pack-

Chung et al. 2007). Virtually all SVs fail to enter axons and accumulate in neuronal cell bodies during the embryonic period. In contrast, mutations in Khc, a catalytic component of kinesin-1, and also mutations in Klc, an accessory component of kinesin-1, cause SV accumulations within axons of larval SGNs (Hurd and Saxton 1996; Gindhart et al. 2003). These studies suggest that, although kinesin-3 is primarily responsible for SV transport, kinesin-1 plays a role in SV transport in a manner distinct from that for kinesin-3, or the two motors may have complementary roles in SV transport at the larval stage. The unc-51 mutant phenotypes, as well as the biochemical evidence that UNC-51 forms a complex with UNC-76/KHC, are most consistent with the idea that UNC-51 functions in SV transport through a kinesin-1dependent pathway. In unc-51 mutants, SVs accumulate within axons at sites distant from cell bodies, suggesting that SVs could partially transport along axons in the absence of unc-51. It is possible that maternally deposited unc-51 contributes to partial transport of SVs into axons, as suggested for defective SV transport in Khc mutants (Hurd and Saxton 1996). It is also possible that there are specific subcellular locations (e.g., axon hillock) or earlier developmental periods where SV transport does not require unc-51 activity. The axonal SV accumulation in unc-51 mutants could be a result of spatially distinct requirement of unc-51 activity for maintaining SV-motor integrity during transport. In addition, as we discussed, kinesin-3 likely contributes to SV transport in unc-51 mutants, resulting in translocation of a subset of SVs into axons and to synapses. SVs that are localized to unc-51 mutant NMJs may reflect a subpopulation of those that were carried by kinesin-3 and did not need unc-51 activity for their transport. In summary, the cooperative action of multiple pathways, including unc-51, unc-76, kinesin-1 and kinesin-3, may be necessary for complete SV transport. Phosphorylation-mediated motor–cargo assembly Previous in vitro studies that addressed a role of phosphorylation in regulating organelle motility have remained unclear and controversial. A series of kinases, including PKA, PKC, and PKG, were shown to have no effect on kinesin-dependent axonal transport (Bloom et al. 1993), whereas phosphorylation of kinesin by PKA was proposed to inhibit fast axonal transport and kinesin binding to membrane organelles (Sato-Yoshitake et al. 1992; Okada et al. 1995). Glycogen synthase kinase 3␤ (GSK3␤) phosphorylates KLC and kinesin-based motility is inhibited by perfusion of active GSK3␤ into squid axons (Morfini et al. 2002). An inhibitory effect of CaMKII in disrupting KIF17–Mint1 association in vitro has recently been reported (Guillaud et al. 2008). However, it was not until recently that the physiological role of phosphorylation in axonal transport was addressed in vivo (Horiuchi et al. 2007), in which JNK-mediated phosphorylation inhibits the kinesin-1–JIP1 adaptor interaction.

GENES & DEVELOPMENT

3301

Toda et al.

Our study has revealed a critical role of UNC-76 phosphorylation during axonal transport, which is likely elicited at the motor–cargo interface. Several lines of evidence support this notion. First, unc-51 genetically interacts with the kinesin-1 adaptor unc-76 in axonal transport in vivo. Second, biochemical experiments show that the association of UNC-76 and Syt-1 is mediated by UNC-51-dependent phosphorylation of UNC-76. Third, FRET analysis confirms the direct association between UNC-76 and Syt-1 in cells, which is attenuated by inhibiting UNC-51 kinase activity. Finally, the SV transport defects of unc-76 mutants are rescued in vivo by phosphomimetic UNC-76, but not by phosphodefective UNC-76. Based on these findings, we propose a model in which adaptor phosphorylation is a key regulatory step that maintains motor–cargo association within axons and leads to efficient SV transport (Fig. 7). Upon phosphorylation by UNC-51 kinase, UNC-76 displays an increased affinity to SV membrane proteins such as Syt-1. Attenuation of UNC-51 kinase activity would reduce the affinity of UNC-76 for SV membrane proteins and cause the

Figure 7. Schematic model for UNC-51 kinase-mediated motor–cargo assembly Our results suggest a model in which UNC51 kinase functions in axonal transport via the kinesin-1-dependent pathway. When phosphorylated by UNC-51 kinase, UNC76, a KHC adaptor and a potential activator of kinesin-1, displays an increased affinity to SV membrane proteins such as Syt-1. UNC-51 likely phosphorylates additional substrates that are essential for axonal transport (dashed arrow). Loss or attenuation of UNC-51 kinase activity would result in a lower affinity of UNC-76 to SV membrane proteins and dissociation of SV cargoes from the motor complexes, as might be the case at NMJ. Kinesin-3 (imac/UNC-104) is also responsible for SV transport in a large fraction and is important for carrying SVs from neuronal soma to axons and synapses (red arrow).

3302

GENES & DEVELOPMENT

dissociation of SV cargoes from the motor complexes. In our model, motor–cargo affinity could also be reduced by dephosphorylation of UNC-76, although such regulatory factor is yet to be identified. Attenuation of UNC-51 activity or activation of phosphatase activity could explain the mechanism of SV cargo detachment from the kinesin motors. Additional work is needed to determine how UNC-51 kinase activity is spatially and temporally controlled to regulate axonal transport. It is notable that both UNC-51 and UNC-76 are undetectable at NMJ of the wild-type third instar larvae (Supplemental Fig. S11), suggesting a spatial control of motor–cargo dissociation at the axonal termini. The proposed mechanism can explain the transport of a subset of SVs. Only ∼20% of SVs appear to colocalize with UNC-76 in wild-type SGNs (Fig. 4D), suggesting that an additional motor/adaptor system, such as kinesin-3, is responsible for carrying the rest of the SVs at the larval stage, as discussed earlier. Indeed, a subpopulation of SVs successfully reaches the synapses in unc-51 mutant NMJs, although these synaptic boutons are smaller in size and fewer in number (Supplemental Fig. S2). It is also likely that UNC-51/UNC-76/kinesin-1 activity is dispensable for loading SVs onto the motor complexes at cell bodies, because SV aggregates are found within axons and distant from cell bodies in these mutants. Thus, the UNC-51/UNC-76/kinesin-1 complex seems important for maintaining motor–cargo association within axons rather than being responsible for initial cargo loading. Alternatively, maternally deposited unc51 or Khc may contribute to partial transport of SVs into axons, as discussed earlier, and the potential role of UNC-51/UNC-76/kinesin-1 complex in initial cargo loading may be masked in our analysis of unc-51 mutants. Although our study clearly demonstrates that phosphorylation of UNC-76 by UNC-51 kinase is critical for SV transport, the phosphomimetic UNC-76 transgene failed to rescue defective SV transport in unc-51 mutants (data not shown). This implies that phosphorylation of UNC-76 is not sufficient for SV transport, and suggests that additional targets of UNC-51 phosphorylation are necessary for proper SV transport. Both KHC and KLC are phosphoproteins (Hollenbeck 1993; Morfini et al. 2002), and unc-51 interacts genetically with Klc. Therefore, additional transport components, including KHC and KLC, need to be tested as candidate substrates for UNC-51 kinase in order to understand the whole picture of unc-51-mediated axonal transport machinery. It is unclear how loss of unc-51 results in aggregation of the kinesin motor complex. Biophysical studies show that cargo binding to the kinesin tail domain is required to unfold kinesin molecules and to activate their motor function on MTs (Coy et al. 1999; Friedman and Vale 1999; Hackney and Stock 2000), suggesting that a failure of motor–cargo assembly could cause stalling and aggregation of kinesin motors. With respect to the kinesin motor activation, a recent report has addressed a novel role for UNC-76/FEZ1 in kinesin motor unfolding and thus activation (Blasius et al. 2007). In this model, two

Regulation of axonal transport by unc-51 kinase

scaffolding/adaptor proteins, JIP1 and UNC-76/FEZ1, induce a step-wise conformational change in kinesin-1, leading to its full activation as a motor in vitro. In good agreement with this model, loss of unc-76 results in disorganized localization of SVs and KHC in vivo (Supplemental Fig. S7). Taken together with our finding that the phosphomimetic UNC-76 is capable of rescuing axonal transport defects in unc-76 mutants, it is possible that UNC-76/FEZ1 not only serves as a motor–cargo linker, but also functions as a kinesin-1 activator in a phosphorylation-dependent manner. Thus, this phosphorylationdependent regulation of adaptors may address an additional mechanism for controlling kinesin-1 activity that is essential for axonal transport. Although the unc-51 and unc-76 pathways could cooperatively activate kinesin-1 activity, it is unlikely that loss of unc-51 leads to complete loss of Khc activity. In unc-51 mutants, mitochondrial transport is partially attenuated, but the majority of mitochondria are still transported, suggesting that the overall activity of KHC, a major motor for mitochondrial transport (Stowers et al. 2002), is preserved. This view is supported by our immunohistochemical evidence, which shows that a subpopulation of KHC is present in aggregates, whereas the rest of KHC was distributed throughout the axons. This suggests that a part of KHC activity may be unaffected in unc-51 mutants, and that the UNC-76-independent population of KHC is functionally active and participate in the transport of other cargoes. Although UNC-51 forms a complex with KHC via UNC-76 and KHC distribution is altered in unc-51 mutants, which strongly suggests a functional interaction between UNC-51 and KHC, it is possible that UNC-51 regulation of kinesin-1 activity is mediated through UNC-76 or an additional factor such as KLC, and the effect of UNC-51 on the KHC motor may be indirect. In this regard, it is suggestive that unc-76 and unc-51 show a clear genetic interaction, whereas Khc and unc-51 do not exhibit an apparent genetic interaction. Implication of unc-51-mediated regulatory process in additional biological contexts Our model of phosphorylation-dependent regulation of motor–cargo assembly could be extended to include additional adaptors or cargo vesicles. UNC-14, a protein that interacts with UNC-51 (Ogura et al. 1997), has recently been reported to play a role in kinesin-1-dependent axonal transport in C. elegans (Sakamoto et al. 2005). UNC-14 might serve as an adaptor for the kinesin motor complex to regulate motor–cargo affinity in an UNC-51-dependent manner. In addition, unc-51 mutations also result in aggregation of vesicles positive for UNC-5, a Netrin/UNC-6 receptor (Ogura and Goshima 2006). Again, affinity between the UNC-5-positive vesicles and their respective motor complexes might be regulated by UNC-51-dependent phosphorylation. Previous studies in worms and mice (Ogura et al. 1994; Tomoda et al. 2004), as well as this study, addressed a role of unc-51 in axon formation. It remains to be studied

whether our model could be extended to explain the regulation of membrane components necessary for axon formation, in which the assembly of axonal membranes with the corresponding motors may be mediated via unc51-dependent phosphorylation. Recent studies identified unc-51 as a homolog of atg1, which plays a role in autophagy (Scott et al. 2004, 2007), a catabolic cellular process responsible for bulk degradation of proteins and organelles, particularly when cells are under nutrient-deprived conditions (Klionsky and Ohsumi 1999). Disruption of the autophagy genes atg5 or atg7 in mouse brains results in neuronal cell death, which accompanies intracellular accumulation of ubiquitin-positive aggregates (Hara et al. 2006; Komatsu et al. 2006). We observed neither ubiquitin-positive aggregates nor symptoms of cellular death in unc-51 mutant SGNs (data not shown), suggesting that the role of unc-51/atg1 in axonal transport is distinct from its role in autophagy, which is induced under nutrient-deficient conditions (Scott et al. 2004). Although autophagy critically depends on intracellular vesicle transport (Klionsky and Ohsumi 1999), and UNC-51 kinase activity seems to be required for autophagy induction (Scott et al. 2007), a link between axonal transport and autophagy remains to be studied. In conclusion, our study identifies a novel regulatory step for axonal transport that depends on the UNC-51 kinase-mediated phosphorylation of a kinesin adaptor. Further studies on the regulation of unc-51 activity will provide us with a better understanding of axonal transport, as well as dynamic neuronal control of synaptic development and plasticity.

Materials and methods Drosophila strains and maintenance Fly strains were maintained on standard medium at 25°C. The following mutants and transgenic fly strains were used: OK6Gal4 (gift from Christoph Schuster), UAS-mito⬋GFP (gift from Bill Saxton; Horiuchi et al. 2005), SydD1 and Syd2H2A (gift from Larry Goldstein), UAS-LAMP1⬋GFP (gift from Helmut Krämer; Pulipparacharuvil et al. 2005), Dhc64CKG08838, Dhc64C4-19, elavc155-Gal4, Khc8, Klc8ex94, unc-76l(1)G0360, unc-76Df(1)107, Aplip1ek4, Aplip1Df(3L)Fpa2, unc-104d11204, Lis-1K13209, UAS-Syt1⬋eGFP, UAS-Rab5⬋YFP (Bloomington Drosophila Stock Center; Zhang et al., 2007; YFP tag at the N⬘ terminus of Rab5), EP(3)3348 (Szeged Drosophila Stock Center), Df(3L)ED4486 (European Drosophila Stock Center). Fly lines carrying transgenes (UAS-unc-51, UAS-unc-51K38A, UAS-Venus⬋unc-51, UAS-unc76, UAS-unc-76S143D, UAS-unc-76S143A) were generated by Genetic Services, Inc., or by Rainbow Transgenic Flies, Inc. Plasmid construction Full-length UNC-51 (LD18893), UNC-76 (LD08195), and KHC (SD02406) cDNAs were obtained from DGRC (Drosophila Genetic Resource Center, Kyoto). To obtain the full-length Syt-1 cDNA, RT–PCR was performed on mRNA purified from wildtype third instar larval brain. The coding regions of these genes were inserted into the SalI/NotI sites of pRK5-HA/myc for expression in mammalian cells, pGEX4T-2 for GST fusion protein

GENES & DEVELOPMENT

3303

Toda et al.

expression in Escherichia coli, pET-28c(+) for 6xHis tag protein expression in E. coli, or into the EcoRI/NotI sites of pUAST for expression in Drosophila. Deletion constructs of UNC-51 and UNC-76 were amplified by PCR. Single amino acid point mutations were introduced into UNC-51 and UNC-76 by site-directed mutagenesis. Syt-1 and UNC-76 were fused in-frame with Cerulean and Venus, blue (CFP) and yellow (YFP) variants of green fluorescent protein (GFP), respectively, with a linker sequence of six glycine residues at the fusion site, and inserted into the SalI/NotI sites of pRK5. Larval tissue staining Larval tissue preparation and staining were done as described in Kurusu et al. (2002). The primary antibodies used were rabbit anti-Syt-1 (1:1,000; gift from Troy Littleton), guinea pig antiUNC-76 (1:100), guinea pig anti-UNC-51 (1:100), mouse antiCSP (1:10; Developmental Studies Hybridoma Bank [DSHB]), mouse anti-FasII (1:10; DSHB), FITC-conjugated HRP (1:100; Jackson ImmunoResearch), rabbit anti-KHC (1:200; Cytoskeleton, Inc.), and mouse anti-Futsch (MAP1B-like; 22C10) (1:5; DSHB). The secondary antibodies used were Alexa 543-conjugated goat anti-rabbit, Alexa 488-conjugated goat anti-rabbit, Alexa 543-conjugated goat anti-guinea pig, and Alexa 633-conjugated goat anti-mouse (1:500; Molecular Probes). To quantify the number of aggregates for CSP, Syt-1, KHC, or UNC-76, background levels were subtracted from each image file by applying a threshold (scaling factor 4) and the particle numbers scored using ImageJ software (NIH). In situ analysis of unc-51 mRNA expression was done as described in Hauptmann and Gerster (2000), using full-length Drosophila unc-51 cDNA. Time-lapse video microscopy Wild-type or unc-51 mutant wandering third instar larvae carrying both UAS-mito⬋GFP and the motor neuron driver OK6Gal4, or those carrying both UAS-Syt-1⬋eGFP and OK6-Gal4 were dissected quickly (